Thanks, miguel :)
Let me explain to you what a dynamic range compressor does and how you can use it.
The compressor might look simple as the very core of it is just four knobs: threshold, ratio, attack, release. But it can be used do to very different things to the sound.
The threshold is always set to a specific signal you're using - so you need to use your ears and meters while setting threshold. Simple rule is the lower the threshold, the more compressor is doing to the sound. If you rise the threshold higher than any peak in the signal - the compressor should do nothing.
Ratio is how steep the compressing curve is. 1:1 means no compression. 4:1 means that a sound that is 4 dB above the threshold will be attenuated so that is stick out only by 1 dB above the threshold. 10:1 is considered a limiter, 20:1 is considered a brickwall limiter. You get it - the higher the ratio, the more compression you get.
Now you have to superimpose timing onto this. And here's where attack and release controls kick it. The attack is the time that the compressor will take to actually reach the desired attenuation (gain reduction level). Say you have a signal that's silence for 2 seconds, then out of a sudden it's a square wave peaking at -10 dB for another 2 seconds. If you've put this through a compressor with threshold of -20 dB, ratio 2:1, attack 200 ms - the square wave would start at -10dBFS and it'd fade down to -15dbFS (10 dB above the threshold divided by 2) by 200 ms. SO the attack is the time needed to turn the compressor's "hidden gain knob" down.
Release - it's the opposite. It's how long it'll take to turn the "hidden gain knob" up, when the input signal drops in loudness.
Please note that compressors might distort the signal if you set the attack and/or release time to short - the compressor will operate so fast, it would change the waveform of the signal creating distortion - it might be desired, but most of the times you don't want that.
Also, compressors frequently feature an RMS/peak control. RMS (Root Mean Square) is the effective electric power of the signal and is always lower than the peak value. It also has a short "window" so it's changing slower than the peak values. If you want sharper, more agressive compression - use peak detection, if you want softer, more transparent compression - go for the RMS detection. Sometimes you can even blend between the two.
The last control you might want to use is makeup gain. It's a simple amplification of the output signal. The (downward) compressor is attenuating parts of the recording that are louder than threshold. So the compressors output is never louder than it's input (unless it's distorting the sound creating harmonics). To compensate that loudness loss - you have the makeup gain, to bring the levels up, after the compressor did it's job.
One thing is you can use it with fast attack and release, and low ratio (1,25:1 through 2:1) to even out the loudness of the signal. If you want to preserve more transients, turn up the attack so the compressors "overlooks" the short impulses, leaving more life and dynamic differences in the signal. If you make the release longer, the loud peaks will "shadow" whatever is after them whitch might be undesired. However - you might want that to attenutae the tail of a snare shot, accenuating the transient - making it more "punchy" and shorter.
Another way of using the compressor is using a very hard compression that is then mixed with the uncompressed signal. This is called "New York compression" or "parallel compression". This helps bring up the quiet detail, and add more power and "meat" to the signal, without loosing too much dynamic range. It's also nice to use this with reverb or delay effects - if you make them quiet enough to be inaudible when the dry (unprocessed) signal is present, they'll fade in right after the dry signal ends if you use a compressor with low threshold (right above the level of the reverb or delay tail), high ratio (4:1, 8:1 etc).
Compression is sometimes used to "glue" the instruments together. Evening out their loudness changes over time lets them sound more consistent together, because they stand in their parts, not standing out and hiding. This is why compression can be used on virtually every track you have in your composition - to make it behave better in relation to others, and to keep it in control.
Finally compression in extreme form (called limiting) is used to prevent signal peaks to rise above the threshold. A typical limiter has less controls, because it's attack and ratio are hard-coded. The shorter the attack, the better - the less signal peaks will be able to stand above the threshold. The ration is typically set to inf:1 - no matter how loud, the signal will be attenuated all the way down to the threshold level.
Limiters also frequently have a "lookahead" option. It;s used to avoid distortion with minimal attack times, and to make the limiting more delicate - the limiter delays the processed signal by a short amount of time (typical value: 5 ms) and compresses the delayed signal using the original (not delayed) to drive the compression. The attack can be increased to the value of the lookahead. So a limiter with lookahead of 5 ms delays the signal by 5 ms and uses attack of 5 ms to smoothen the process. The problem with that is the increased latency (using such a limiter in parallel mode would add a comb filter-like effects). Also the "pre-atennuation" might be undesired in some situations. A few limiters allow you to change the lookahead.
Posted by: unfa on 2014-10-18 14:18:58